Control system and control method

Abstract

To control a plant according to a change in the characteristics of the plant to be controlled.SOLUTION: A control model is stored in a memory for each characteristic of a plant, the control model defining an action satisfying a predetermined condition among actions in a state transition model in which a second state transitioned by an action in a first state is defined for each characteristic of the plant, a processor determines a state of the plant in the characteristic based on operation data obtained from the plant and the state transition model, and outputs operation information for performing an operation on the plant in the state based on the determined state of the plant in the characteristic and the control model.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a control system and a control method. [Background technology]

[0002] In recent years, there has been a growing need to apply AI technology to the industrial sector, and the number of cases in which AI-based control technology is being applied is also increasing in the field of process control.

[0003] For example, Patent Document 1 discloses an example of an extension of a future state estimation device to the optimization of model-based control, which device includes: a model storage unit that stores a state transition model in which the future states of a simulated object and its surrounding environment at an infinite time or time steps ahead in a finite space are expressed in the form of a probability density distribution using the probability of transition from a source state to a destination state; a future state prediction calculation unit that performs calculations to determine the future states of the simulated object and its surrounding environment using the state transition model; and a future state prediction result storage unit that stores information determined by the future state prediction calculation unit, wherein the state transition model determines a decaying state transition matrix D representing the future state as D=T(E-γT)-1 from a state transition probability matrix T, a unit matrix E, and a decay rate γ as a weight. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2019-159876 Summary of the Invention [Problem to be solved by the invention]

[0005] According to Patent Document 1, the future state of a control object and its surrounding environment in an infinite time future can be calculated in the form of a probability density distribution, regardless of the time until the future state to be predicted, and by using this calculation result, an optimal control law can be calculated taking into account the future state in an infinite time future. Therefore, for example, in process control, if it is desired to switch from one state A (e.g., temperature TA) to another state B (temperature TB) early, the optimal operating conditions for transitioning from state A to state B can be found in a short time.

[0006] However, the characteristics of a plant do not always remain the same. For example, the characteristics of a plant may change due to deterioration or failure of the equipment that constitutes the plant. Furthermore, the characteristics of a plant may change due to an abnormality that occurs in the plant. Conventionally, when the characteristics of the plant to be controlled, including the characteristics of the equipment that constitutes the plant, change, it has been difficult to operate the plant appropriately in response to the change in the characteristics. However, Patent Document 1 does not take such a problem into consideration.

[0007] An object of the present invention is to provide a technique that is capable of controlling a plant in accordance with changes in the characteristics of the plant to be controlled. [Means for solving the problem]

[0008] The control system according to the present invention is a control system that controls a plant to be controlled by a computer having a processor and a memory, wherein the computer stores in the memory, for each characteristic of the plant, a control model that defines actions that satisfy predetermined conditions among actions in a state transition model that defines a second state to which a transition occurs due to an action in a first state for each characteristic of the plant, and the processor determines the state of the plant for the characteristics based on operating data obtained from the plant and the state transition model, and outputs operation information for operating the plant in the state based on the determined state of the plant for the characteristics and the control model. [Effects of the Invention]

[0009] According to the present invention, it is possible to control a plant in accordance with changes in the characteristics of the plant to be controlled. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 2 is a diagram showing the configuration (control system) during offline learning in this embodiment. [Figure 2] FIG. 2 is a diagram illustrating a processing flow of the present embodiment. [Figure 3] FIG. 1 is a diagram illustrating an example of a schematic configuration of a computer. [Figure 4] FIG. 1 is a diagram illustrating an overview of a plant. [Figure 5] FIG. 4 is a diagram illustrating an example of operation data. [Figure 6] FIG. 1 is a diagram illustrating an overview of a state definition of a plant. [Figure 7] FIG. 1 is a diagram illustrating an overview of a state transition model. [Figure 8] FIG. 2 is a diagram illustrating details of a state transition model. [Figure 9] FIG. 10 is a diagram illustrating an example of a control model. [Figure 10] FIG. 10 is a diagram illustrating a control model in comparison with a state transition model. [Figure 11] 10A and 10B are diagrams illustrating examples of evaluation results by a control performance evaluation unit. [Figure 12] FIG. 10 is a diagram illustrating an example of state transitions representing results of the control performance evaluation unit. [Figure 13] FIG. 10 is a diagram showing updated state transitions. [Figure 14] FIG. 10 is a diagram illustrating an example of a controllability information display unit. [Figure 15] FIG. 1 is a diagram showing a configuration (online control system) during online control in this embodiment. [Figure 16] FIG. 1 is a diagram illustrating an example of the configuration of a system in which a control system and an online control system are linked together. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments of the present invention will be described with reference to the drawings. The examples are illustrative for explaining the present invention, and appropriate omissions and simplifications have been made for clarity of explanation. The present invention can be implemented in various other forms. Unless otherwise specified, each component may be singular or plural. The position, size, shape, range, etc. of each component shown in the drawings may not represent the actual position, size, shape, range, etc., in order to facilitate understanding of the invention. Therefore, the present invention is not necessarily limited to the position, size, shape, range, etc. disclosed in the drawings.

[0012] Examples of various types of information may be described using expressions such as "table," "list," and "queue," but the various types of information may also be expressed using data structures other than these. For example, various types of information such as "XX table," "XX list," and "XX queue" may also be expressed as "XX information." When describing identification information, expressions such as "identification information," "identifier," "name," "ID," and "number" are used, but these are interchangeable.

[0013] When there are multiple components with the same or similar functions, they may be described using the same reference numeral with different subscripts. When there is no need to distinguish between these multiple components, the subscripts may be omitted.

[0014] In the embodiments, processing performed by executing a program may be described. Here, a computer executes the program using a processor (e.g., a CPU or a GPU) and performs processing defined by the program using storage resources (e.g., a memory) and interface devices (e.g., a communication port). Therefore, the entity performing the processing by executing the program may be the processor. Similarly, the entity performing the processing by executing the program may be a controller, device, system, computer, or node having a processor. The entity performing the processing by executing the program may be any computing unit, and may include a dedicated circuit that performs specific processing. Here, the dedicated circuit may be, for example, an FPGA (Field Programmable Gate Array), an ASIC (Application Specific Integrated Circuit), or a CPLD (Complex Programmable Logic Device).

[0015] A program may be installed on a computer from a program source. The program source may be, for example, a program distribution server or a computer-readable storage medium. When the program source is a program distribution server, the program distribution server may include a processor and a storage resource for storing the program to be distributed, and the processor of the program distribution server may distribute the program to be distributed to other computers. In addition, in an embodiment, two or more programs may be realized as one program, or one program may be realized as two or more programs.

[0016] FIG. 1 is a diagram illustrating an example of the configuration of a control system according to this embodiment. The control system 1000 according to this embodiment is a system for constructing, offline in advance, a control algorithm, a state transition model, and a control model for a control device for controlling a plant. The control system 1000 includes a plant 1, a simulator 2, a condition setting unit 3, operation data 4, a control device 5, a control performance evaluation unit 6, a control feasibility information adding unit 7, control feasibility information 8, and a control feasibility information display unit 9. Although the plant 1 and the simulator 2 are included in this system in FIG. 1, they may also be configured as separate systems. That is, the control system 1000 may construct a control algorithm, a state transition model, and a control model using operation data 4 obtained from the plant 1 and the simulator 2 as input.

[0017] The plant 1 is a controlled object, and in this embodiment, it is a batch plant. The simulator 2 is a dynamic characteristic simulator for the controlled batch plant and reproduces the behavior of the batch plant. The condition setting unit 3 is a processing unit that sets the operating conditions of the simulator 2, specifically, the operating variables such as valve openings and the temperature setpoints manually operated by a human. The operating data 4 includes actual operating data output from the plant 1 and simulated operating data created by the simulator 2. The operating data 4 also includes, as attribute data, a label indicating the plant state, such as normal or abnormal, indicating the type of state of the plant 1. As will be described later in detail, the label is information indicating the plant state, and controllability information 8 is added as information indicating whether the plant 1 in the plant state to which the label is assigned is controllable. Hereinafter, the "plant state" may also be referred to as a "state," but is referred to as a "plant state" to distinguish it from a state in a state transition model.

[0018] The control device 5 includes a state determination unit 51, a state transition model 52, and a control model 53. The state determination unit 51 uses data clustering technology to learn from the operation data 4, thereby defining a plant state for each plant condition so as to distinguish between normal operation data and abnormal operation data. The state determination unit 51 creates a state transition model 52 and a control model 53 for each plant characteristic determined according to the defined plant state. Specifically, as will be described later, the state determination unit 51 acquires operation data, determines the characteristics of the plant to be analyzed based on labels corresponding to the plant state of the operation data, and creates a state transition model 52 and a control model 53 that define the relationship between states in the plant for each determined plant characteristic. The state transition model 52 is a model that defines, using a probability matrix, the states to which a certain action will transition from a certain state. The control model 53 is a model that defines the optimal action for ultimately reaching a certain state B from a certain state A in the state transition model 52. Here, the optimal action refers, for example, to being able to reach state B with the fewest operations.

[0019] The state transition model 52 and the control model 53 are stored for each plant characteristic determined by the state determination unit 51. Here, the plant characteristics are determined based on the plant state (e.g., based on the type of plant state) as described above, and can be understood as the plant behavior (e.g., plant behavior caused by the characteristics of the equipment constituting the plant) determined based on the type of plant state in a certain state. Therefore, a change in the plant characteristics occurs, for example, due to a change in the behavior of the equipment constituting the plant or a change in the plant behavior caused by an abnormality occurring in the plant. The state transition model 52 and the control model 53 are stored for each plant characteristic corresponding to the type of plant state, such as normal or abnormal, and / or the type of abnormality when the type is abnormal. In this way, it can be said that the plant characteristics are determined based on the plant state occurring in the plant. As will be described later, an abnormality occurring in the plant may include an abnormality related to plant deterioration.

[0020] As will be described in detail below, in this embodiment, the characteristics of such a plant are determined based on the type of plant state that changes depending on the behavior of the equipment that makes up the plant, and, if the type is abnormal, the type and degree of the abnormality.Even if the plant characteristics obtained by this determination change, the plant is controlled in accordance with the change in the characteristics.

[0021] Examples of the behavior of the equipment constituting the plant include the following. For example, various operations occur due to differences in the operating status or maintenance status of the equipment, and various operations according to the operating mode for changing the plant output depending on environmental differences such as temperature and humidity. Specifically, examples include operations according to the operating status of the equipment, such as differences in the rotation speed of the agitator blades 13, and maintenance status, such as differences in the number of times the valve 15 has been replaced.

[0022] Furthermore, when the type of plant state is abnormal, the behavior for determining the type and degree of the abnormality may be the type and degree of the abnormality that has occurred in the plant (or the equipment that constitutes the plant). Examples of such behavior include the following. For example, equipment operation according to the type and degree of the abnormality may be exemplified. Specifically, one example is operation according to the type of abnormality, such as insufficient strength of the valve 15, or the degree of the insufficient strength. Even if the type of abnormality is the same, different behaviors may be determined depending on the degree. For example, since the plant characteristics differ depending on the degree of insufficient strength of the valve 15, if the degree of insufficient strength differs, the two may be determined to be different behaviors. In addition, even if the plant is normal, for example, a case where an abnormality has occurred in an equipment that constitutes the plant and a case where no abnormality has occurred in the equipment may be considered to have different plant characteristics, and the two may be determined to be plant behaviors of different types and degrees.

[0023] The control performance evaluation unit 6 links the control model 53 with the simulator 2 and controls the simulator 2 using the control model 53. The control performance evaluation unit 6 evaluates whether the plant can be controlled to satisfy predetermined constraints using the control model 53. For example, the control performance evaluation unit 6 evaluates the control performance of the plant using an index indicating whether constraints such as the time to reach a control target or temperature can be met.

[0024] The controllability information adding unit 7 adds controllability information 8 indicating whether control is possible or not to each state (see FIG. 7 ) defined in the state transition model 52, based on the information of the state transition model 52 and the evaluation result by the control performance evaluation unit 6. The controllability information display unit 9 displays the controllability information 8.

[0025] Next, the processing flow of this embodiment will be explained with reference to FIG.

[0026] In step 101, the state determination unit 51 of the control device 5 collects actual operation data of the plant 1 to be controlled.

[0027] In step 102, the simulator 2 generates simulation operation data as needed. As will be described later in detail, the simulation operation data is generated in accordance with the operation conditions set by the condition setting unit 3 after the control performance evaluation unit 6 evaluates the control performance.

[0028] In step 103, the state determination unit 51 reads the measurement values ​​included in the actual operation data collected in step 101 and / or the pseudo-measurement values ​​obtained by simulation included in the simulated operation data created in step 102, and assigns labels to these operation data. Here, the labels are attribute data that represent the plant state, including the type of plant state (normal or abnormal) of the equipment that constitutes the plant, and the type and degree of abnormality if the type is abnormal. By the processing of step 103, operation data to which the above labels corresponding to the plant state are assigned are created, as shown in FIG. 5, which will be described later.

[0029] In step 104, the state determination unit 51 classifies the operation data by a clustering technique using the labels assigned in step 103. Furthermore, the state determination unit 51 determines the characteristics of the plant being analyzed based on the labels corresponding to the plant states of the classified operation data. For example, the state determination unit 51 determines that the type of plant state (e.g., the type obtained from the label of the operation data classified in s1 in FIG. 6) in a certain state (e.g., s1 in FIG. 6) represented as a set of the classified operation data has different characteristics between a plant in a "normal" state and a plant in an "abnormal" state. Through the processing of step 104, the classified operation data (e.g., the black dots in FIG. 6) are associated with the plant state (e.g., s1 in FIG. 6) and the plant state attribute (the type indicating whether the plant is normal or abnormal) in the plant state, and then classified, thereby determining the characteristics of the plant.

[0030] In step 105, the state determination unit 51 creates a state transition model 52 of the plant having the determined characteristics, for example, based on the temporal transition of the operation data classified in step 104. By performing step 105, the state transition model 52 for the determined characteristics is created.

[0031] Here, in step 105, if the control device 5 has a state transition model 52 created in advance for each characteristic and the state transition model 52 does not need to be updated, the state transition model 52 of the corresponding characteristic may be identified and acquired based on the characteristic of the plant 1 determined in step 104. Furthermore, the state determination unit 51 can determine, through the processing of steps 104 and 105, which state the plant operating based on the acquired operation data is in in the state transition model of the corresponding characteristic. In other words, through the processing of steps 104 and 105, the state of the plant for the characteristic is determined based on the operation data acquired from the plant and the state transition model of the characteristic, thereby making it possible to determine what state the plant for that characteristic is in.

[0032] In step 106, the controllability information adding unit 8 determines whether each "state" (see FIG. 7 ) of the state transition model 52 in the characteristics determined in step 105 satisfies a predetermined constraint condition, and adds controllability information 8 indicating whether the plant 1 is controllable, based on the result of the determination. The controllability information 8 makes it possible to determine whether the plant 1 is controllable simply by referring to the state transition model 52. That is, by performing step 106, it becomes possible to determine whether the plant 1 is controllable in each state (see FIG. 7 ) to which a label indicating whether the plant is normal or abnormal is added. For example, not only is the controllability information 8 added when the plant state type is normal, but also, when the plant state type is abnormal but satisfies a predetermined constraint condition, the controllability information 8 may be added to indicate that the plant is controllable. Furthermore, the determination of whether the predetermined constraint condition is satisfied may be performed using the result of a control performance evaluation performed by the control performance evaluation unit 6, which will be described later.

[0033] In step 107, state determination unit 51 creates control models 53 for each of state transition models 52 created in step 105. A method for creating control models 53 will be described later with reference to FIG.

[0034] In step 108, the control performance evaluation unit 6 uses the control model 53 and the simulator 2 to evaluate the control performance of the plant 1, such as whether it can meet specified constraints such as the time to reach the control target and the temperature, and outputs the evaluation results.

[0035] In step 109, if the evaluation result in step 108 shows that the plant's control performance does not satisfy a predetermined constraint, the control performance evaluation unit 6 determines whether or not the operation can be improved so as to satisfy the predetermined constraint. If the control performance evaluation unit 6 determines that the operation can be improved (step 109; Yes), the process proceeds to step 110; if it determines that the operation cannot be improved (step 109; No), the process proceeds to step 111. Regarding whether or not the operation can be improved, for example, the control performance evaluation unit 6 determines that the operation can be improved so as to satisfy the predetermined constraint if the jacket inlet temperature included in the operating data has a certain margin up to the upper limit value that satisfies the conditions for the plant. Then, the control performance evaluation unit 6 outputs an instruction to the condition setting unit 3 to control the opening and closing of the valve so as to increase the jacket inlet temperature by a predetermined amount.

[0036] In step 110, the condition setting unit 3 creates information for setting operation conditions that are thought to be improvable in accordance with the instruction received from the control performance evaluation unit 6 in step 109, and then proceeds to step 102. Creating the information for setting the operation conditions means, for example, setting, as a condition, an operation to increase the opening / closing angle of the valve stopper from angle "a" to angle "b" so as to approach the upper limit value and satisfy the predetermined constraint condition.

[0037] In step 111, the controllability information adding unit 7 adds the controllability information 8 in association with each "state" of the state transition model 52 (see Figure 7) based on the results of the control performance evaluation performed by the control performance evaluation unit 6.

[0038] In step 112 , the controllability information display unit 9 displays the controllability information 8 added in step 111 .

[0039] Through the above steps, a control model can be constructed that can perform operation in response to changes in the characteristics of the plant to be controlled when the characteristics change.

[0040] The control device 5 shown in Figure 1 can be realized, for example, by a general computer 1600, as shown in Figure 3 (schematic diagram of a computer), which includes a CPU 1601, a memory 1602, an external storage device 1603 such as an HDD (Hard Disk Drive), a reading / writing device 1607 that reads and writes information from a portable storage medium 1608 such as a CD (Compact Disk) or USB memory, an input device 1606 that accepts input of various information such as a keyboard and a mouse, an output device 1605 such as a display that outputs various information that has been input and is used for processing, a communication device 1604 such as an NIC (Network Interface Card) for connecting to a communication network, and an internal communication line (referred to as a system bus) 1609 such as a system bus that connects these.

[0041] Furthermore, various data (e.g., model 34) stored in the control device 3 or used for processing can be realized by the CPU 1601 reading and using the data from the memory 1602 or the external storage device 1603. Furthermore, each unit of the control device 5 (e.g., state identification unit 51, state transition model 52, and control model 53) can be realized by the CPU 1601 loading a predetermined program stored in the external storage device 1603 into the memory 1602 and executing it.

[0042] The above-mentioned predetermined programs and data may be stored (downloaded) into the external storage device 1603 from the storage medium 1608 via the reading / writing device 1607 or from the network via the communication device 1604, and then loaded onto the memory 1602 and executed by the CPU 1601. Alternatively, the programs and data may be directly loaded onto the memory 1602 from the storage medium 1608 via the reading / writing device 1607 or from the network via the communication device 1604, and then executed by the CPU 1601.

[0043] In the following, we will explain an example in which the control device 5 is configured by a single computer, but all or part of these functions may be distributed across one or more computers, such as a cloud, and similar functions may be realized by communicating with each other via a network.

[0044] Examples of the present invention will be described in detail below.

[0045] FIG. 4 is a diagram showing an overview of the plant 1 that is the object of control in this example. The plant 1 is a batch plant. Its main components are a reaction tank 11, a jacket 12, an agitator 13, a pump 14, a valve 15, an existing control device 16, and a temperature controller 17. Raw materials are charged into the reaction tank 11, and a reaction of the raw materials is initiated by an initiator. During operation, the charged raw materials are agitated by the agitator 13 so as to be as uniform as possible. In addition, temperature-adjusted water is sent by a pump 14 to a temperature-adjusting jacket 12 attached to the reaction tank 11, and the temperature Tr measured in the reaction tank is controlled to a target temperature.

[0046] Next, a method for adjusting the temperature Tr of the reactor will be described in detail. In this embodiment, the temperature Tr in the reactor is controlled by changing the inlet temperature Tci of the jacket 12. Specifically, the measured temperature Tr is input to the existing control device 16, which compares the measured temperature Tr with the target value of the temperature Tr stored in the control device 5 and provides a set value (SV) of the jacket inlet temperature Tci to the temperature regulator 17. For example, if the measured temperature Tr is higher than the target value, the control device 5 adjusts the set value of the jacket inlet temperature Tci downward. On the other hand, if the measured temperature Tr is lower than the target value, the control device 5 adjusts the set value of the jacket inlet temperature Tci upward. The temperature regulator 17 compares the set value of the jacket inlet temperature Tc with the actual value of the jacket inlet temperature Tci measured by a temperature sensor and adjusts the opening of the valves 15a and 15b so that the jacket inlet temperature Tci approaches the set value.

[0047] Simulator 2 is a dynamic characteristic simulator for the batch plant to be controlled, and reproduces the behavior of the batch plant. Simulator 2 is built based on a first-principles model, and the relationship between the concentration and temperature of the products in the reaction vessel is described by a differential equation. For example, consider the case where product C is produced from raw materials A and B according to the following reaction formula. A+B→C

[0048] In the above reaction formula, the reaction rate r is generally expressed as follows: r = k[A][B]

[0049] where k is the reaction rate constant, and [A] and [B] are the concentrations of reactants A and B, respectively. This reaction rate equation shows that the reaction rate increases in proportion to the product of the concentrations of reactants A and B. Therefore, the mass balance for product C can be expressed as the following differential equation: d[C] / dt = r

[0050] Here, d[C] / dt represents the rate of change of the concentration of product C over time, and r is the reaction rate. This differential equation shows how the concentration of product C changes as a function of time. By calculating the reaction rate based on the reaction rate equation and substituting it into the differential equation, we can simulate the change in the concentration of product C over time.

[0051] The temperature Tr in the reaction vessel is expressed as follows using the reaction heat Q1 and the amount of cooling by the jacket Q2: dTr / dt = (Q1 - Q2) / (ρCpV) Here, Tr is the temperature inside the reaction vessel, Q1 is the heat of reaction, Q2 is the amount of cooling by the jacket, ρ is the density of the reactants, Cp is the specific heat capacity of the reactants, and V is the volume of the reaction vessel.

[0052] Furthermore, Q2 is expressed by the following equation using the temperature Tr of the reaction vessel and the representative temperature (cooling water temperature) Tc of the cooling water flowing in the jacket. Q2 = U × A × (Tr-Tc) Here, U is the overall heat transfer coefficient and A is the heat transfer area.

[0053] Simulator 2 reproduces the behavior of a batch plant by solving differential equations that take into account such temperature changes.

[0054] The condition setting unit 3 is a processing unit that sets the operating conditions of the simulator 2. Specifically, it sets the initial state of the batch process, the characteristics of various equipment, and subsequent manipulated variables. By appropriately setting the operating conditions of the simulator 2 using the condition setting unit 3, the simulator 2 can simulate a plant with any desired characteristics. The initial state includes the amount of raw material charged, and the equipment characteristics include the volume V of the reactor and the overall heat transfer coefficient U, which accounts for the heat transfer characteristics. The manipulated variables include the valve opening and a set value manually operated by a human. In this embodiment, the manipulated variable is, for example, the target value Trs of the reactor temperature Tr. For example, if the target temperature is 60°C, the target value Trs is normally set to 60°C. Therefore, if the temperature rises, the jacket inlet temperature Tci is lowered by PID control, and Tr is returned to 60°C. However, if for some reason the reactor temperature Tr rises significantly or if it is desired to return the reactor temperature Tr to 60°C more quickly, the target value Trs can be set to, for example, 58°C, thereby lowering the jacket inlet temperature Tci more than usual and enabling the reactor temperature Tr to be lowered more quickly.

[0055] The operation data 4 includes actual operation data output from the plant and pseudo operation data created by a simulator. An example of the operation data 4 is shown in Fig. 5. As shown in Fig. 5, the operation data 4 stores a time indicating the operation time of the plant, the reactor temperature Tr, the jacket inlet temperature Tci, a cooling water flow rate Fc indicating the flow rate of cooling water to the plant, and a label indicating the plant state, all of which are associated with each other.

[0056] FIG. 5 shows, for example, that the control device 5 received data from the plant as operating data at "2023 / 1 / 1 10:01:00" including the reactor temperature Tr of "60.1," the jacket inlet temperature Tci of "35.0," and the cooling water flow rate Fc of "12.1." The data also shows that the type of plant state at that time is "normal," indicating that the control device 5 received operating data including such attribute data from the plant. The data also shows that the control device 5 received data from the plant as operating data at "2023 / 9 / 1 9:01:00" including the reactor temperature Tr of "59.9," the jacket inlet temperature Tci of "30.8," and the cooling water flow rate Fc of "12.2." The data also shows that the type of plant state at that time is "abnormal (small transmission surface degradation)." As described above, the simulated operating data obtained by the simulator 2 through simulation is also stored as operating data 4.

[0057] In this way, a label indicating a type of the plant, i.e., whether it is normal or abnormal, is assigned to the operation data as attribute data. As already explained, the label is assigned based on the measurement values ​​included in the actual operation data collected in step 101 and the pseudo-measurement values ​​included in the simulated operation data created in step 102. For example, if these measurement values ​​contain a code indicating that the plant is normal (normal code), the label "normal" is assigned, and if these measurement values ​​contain a code indicating that the plant is abnormal (error code), the label "abnormal" is assigned. Furthermore, a label including the type of abnormality may be assigned depending on the value of the error code. In FIG. 5, as an example, it can be seen that a label indicating "small transmission surface degradation" is assigned.

[0058] The abnormality refers to a plant state that differs from the normal plant state and includes a deteriorated state of the plant or equipment. Examples of such deteriorated states include a stuck valve and deterioration of the transfer surface. In this embodiment, two abnormal states are defined: a state in which the degree of deterioration of the transfer surface is smaller than a predetermined standard (small surface deterioration), and a state in which the degree of deterioration of the transfer surface is larger than a predetermined standard (large surface deterioration). The simulator 2 generates operating data in which the transfer surface is deteriorated by performing a simulation with a small value of the overall heat transfer coefficient U. Furthermore, when assigning a label to the actual operating data, the state determination unit 51 estimates the value of the overall heat transfer coefficient U from the actual operating data and assigns a label of either "normal" or "abnormal" to the actual operating data depending on the estimated value of the overall heat transfer coefficient U.

[0059] The control device 5 has a state determination unit 51, a state transition model 52, and a control model 53. First, the state determination unit 51 will be described.

[0060] The state determination unit 51 uses data clustering technology to define the state of the plant using the labeled operation data 4 so that the type representing the plant state, such as normal or abnormal, can be determined. The state determination unit 51 defines such a plant "state" as a combination of multiple measurement values ​​included in the labeled operation data, and uses this definition to determine the plant state in each state of the plant. That is, the state determination unit 51 classifies the labeled operation data 4 created by the plant 1 and / or the simulator 2 using data clustering technology, determines the characteristics of the plant to be analyzed based on the plant state type in each state of the plant in the classified operation data, and creates a state transition model 52 of the determined characteristics. This makes it possible to determine the plant characteristics and the type of plant state each state of the plant with the characteristics corresponds to.

[0061] For example, Figure 6 shows an example in which a state is defined by the reactor temperature Tr and its rate of change dTr. The black circles in this figure represent operational data, and the dashed circles represent the plant state. For example, s1 in the figure represents a state in which the reactor temperature Tr is low and the rate of change dTr is large, while s2 represents a state in which the reactor temperature Tr is high and the rate of change dTr is small. In this example, the classification of this state, i.e., the position of the dashed circles, is determined using data clustering technology. Note that in the example of Figure 6, for simplicity, the plant state was explained using two variables, the reactor temperature Tr and the rate of change dTr of the temperature. However, in this example, a state definition is used that uses the jacket inlet temperature Tci in addition to the reactor temperature Tr and the rate of change dTr of the temperature. This is because it is difficult to distinguish between normal and abnormal conditions using only the reactor temperature Tr and the rate of change dTr of the temperature.

[0062] For example, a steady state where the reactor temperature Tr is the target value and the temperature change rate dTr is close to zero is possible under both normal and abnormal conditions. However, in the case of an abnormality in the deterioration of the transfer surface, which is the subject of this example, the plant state changes due to the relationship between the jacket inlet temperature Tci and the reactor temperature Tr required to maintain the steady state. Therefore, the use of three variables makes it possible to distinguish between normal and abnormal conditions. The data clustering technique used to distinguish between these conditions is supervised data clustering, and this technique makes it possible to distinguish between normal, minor, and major transfer surface deterioration conditions.

[0063] Next, the state transition model 52 will be explained. The state transition model 52 defines, for example, the state to which a certain action transitions from a certain state, using a probability matrix. An image of the state transition model 52 is shown in Figure 7. In the example shown in this figure, performing operation a1 from state s0 causes a transition to state s1, and performing operation a2 causes a transition to state s2. Note that this figure is a simplified diagram, and performing an operation in a state before the transition determines one transition destination. However, in reality, performing the same operation can cause a transition to different states, and each has its own transition probability. An actual image of this transition probability is shown in Figure 8.

[0064] In the example of FIG. 8, when operation a2 is performed from state s0, the probability of transitioning to state s1 in the next step is 20%, the probability of transitioning to state s2 is 70%, and the probability of transitioning to state s3 is 10%. Furthermore, when operation a3 is performed from the same state s0, the probability of transitioning to state s5 is 55%, and the probability of transitioning to state s6 is 45%. This state transition model can be obtained by discretizing the operational data into "states" and then statistically processing the relationship between the operation (the set value of the reactor temperature Tr) and the state of the next step. This state transition model is stored for each type, such as when the plant is normal or when an abnormality occurs, and / or for each type of abnormality when the type is abnormal. In this way, the state transition model in this embodiment is stored for each plant characteristic.

[0065] Next, the control model 53 will be described. The control model 53 is a model that defines an action (for example, an optimal action) that satisfies a predetermined condition for finally reaching a certain state sj from a certain state si. The control model 53 can be obtained from the state transition model 52. In this embodiment, the control model 53 is obtained by applying a reinforcement learning algorithm. That is, the control model is determined by determining the optimal action by giving a large reward when taking a certain action from a certain state makes it easier to approach the final target state. Note that the optimal route for transitioning from a certain state to the final target state may also be obtained by dynamic programming.

[0066] FIG. 9 shows a portion of an example of a control model. This example shows a case where the final target state is s10. In this example, when the state is S0, taking operation a1 is optimal, resulting in a transition to state s1. Furthermore, when the original state is s1, the optimal operation is a4, resulting in a transition to state s3. Adding this information to the state transition model of FIG. 7 results in FIG. 10. Here, the target state is state s10, and the control model is shown by the solid line. For example, when the current state is state s0, the candidate operations are operation a1, operation a2, and operation a3, but the optimal operation is operation a1, shown by the solid line. Furthermore, when the current state is state s2, the optimal operation is operation a2. In this way, in the control model 53, as an example, actions that quickly reach a specified state are stored in association with states with high transition probabilities.

[0067] The control performance evaluation unit 6 links the control model 53 with the simulator 2, controls the simulator 2 using the control model 53, and generates control performance information, which is information related to the evaluation of the control process based on the control model 53. The control performance information is information obtained by determining whether indicators related to the plant's control process satisfy predetermined criteria. Examples of indicators related to the plant's control process include whether a target can be reached, the time required to reach the target, and parameters such as temperature and pressure. The control performance is evaluated based on information such as whether the time required to reach the target satisfies predetermined criteria and whether parameters such as temperature and pressure exceed or fall below predetermined thresholds. Note that the control performance information is not limited to this example. For example, other information indicating the stability and operability of the control process or other information indicating compliance with constraints that must be observed can be used as the control performance information.

[0068] In this example, control performance was evaluated based on whether temperature constraints could be met. Two upper temperature constraints were established: temperature T1 and temperature T2. The constraint was that temperature T1 must be less than temperature T2, and temperature T2 must not be exceeded. Temperature T1 must not be exceeded continuously for a certain period of time Δt1 or longer. Figure 11 shows an example of control performance evaluation. The graph in the figure shows the transition of reactor temperature Tc, with time on the horizontal axis. If the results output by control model 53 are shown as solid line 1101, the constraints of temperature T1 and temperature T2 are met. On the other hand, if operation is performed under different conditions, such as exceeding temperature T1 for a period longer than Δt1 (as shown by dashed line 1102) or reaching a temperature exceeding temperature T2 (as shown by dashed line 1103), the operation is deemed inappropriate. For example, if the control performance does not meet the specified conditions, such as if the constraints cannot be met, the operating conditions are set in step 110 shown in Figure 2. Then, in step 102, the simulator 2 creates the simulation operation data, and the subsequent processes are carried out to create the state transition model 52 and the control model 53 that reflect the new operating conditions.

[0069] When the control model is updated by creating the new control model, the simulator 2 is then controlled by the updated control model 53, and the control performance in that case is evaluated. The process shown in Fig. 2 is repeated until it is determined that the control performance satisfies a predetermined standard, or until there are no more candidates for the operation to be evaluated in step 109. The specific procedure will be described with reference to Fig. 12.

[0070] Fig. 12 is a diagram schematically showing state transitions obtained from the evaluation results of the control performance described with reference to Fig. 11. For the sake of explanation, state numbers in the 10s represent states where the temperature is lower than T1, state numbers in the 20s represent states where the temperature is between T1 and T2, and state numbers in the 30s represent states where the temperature is equal to or higher than T2. ​​There are five types of operations, a1 to a5, and the order of effect in lowering the temperature is operation a5 > operation a4 > operation a3 > operation a2 > operation a1.

[0071] 12, the state transition to state s31, where the temperature is equal to or greater than the limit value T2, is the result of performing operation a5 from state s21. Operation a5 is the operation that has the greatest effect of lowering the temperature (i.e., in this example, the effect of quickly reaching the target value), and is the best operating condition. Therefore, the control device 5 passes this information to the controllability information adding unit 7. On the other hand, the state transition when the temperature remains at the limit temperature T1 for a long time is state s12 → state s22 → state s23 → state s18. Here, the operation that occurred when the state transitioned from state s12 to state s22 was operation a4, which is not the best operation a5. Therefore, the control device 5 passes the operating condition for performing operation a5 from the state represented by state s12 to the condition setting unit 3. By repeating each step shown in FIG. 2, the result of performing the best operation is obtained.

[0072] Next, the controllability information adding unit 7 will be described. Based on the information of the state transition model 52 and the evaluation results of the control performance evaluation unit 6, the controllability information adding unit 7 adds controllability information 8 indicating whether or not control is possible to each state defined in the state transition model. In the example of FIG. 13 in which the information of FIG. 12 is updated, the controllability information adding unit 7 labels states s31 and s32, which exceed the temperature T2 that is the constraint condition, as "uncontrollable." Furthermore, because the control performance evaluation unit 6 output an evaluation result indicating that the state transition from state s21 to state s31 is the best operation, it can be seen that a transition to state s31 will occur when state s21 is reached. Therefore, the controllability information adding unit 7 also labels state s21 before reaching state s31, which actually becomes uncontrollable, as "uncontrollable." In other words, the controllability information adding unit 7 determines whether the evaluation result of the state to transition to next after the state to be labeled is the evaluation result of a state to which the "uncontrollable" label should be assigned. Then, if the controllability information adding unit 7 determines, as a result of the judgment, that the evaluation result of the next transition state is the evaluation result of a state to which the label "uncontrollable" is to be assigned, it also assigns the label "uncontrollable" to the state to which the label is to be assigned.

[0073] On the other hand, in the example where the state changes from state s12 to state s22 to state s23 to state s18, performing operation a5 from state s12 causes a transition to state s23, and continuing operation a5 causes a return to state s18. This type of state transition does not fall under the constraint that the limit temperature T1 must not be exceeded continuously for a certain period of time Δt1 or more. Therefore, the controllability information adding unit 7 assigns the label "controllable" to state s12. As described above, the controllability information adding unit 7 assigns the label "uncontrollable" or "controllable" to each state in this way.

[0074] Finally, the controllability information display unit 9 will be described. An example of the controllability information display unit 9 is shown in FIG. 14. In this example, the controllability information display unit 9 maps states with the horizontal axis representing the jacket inlet temperature and the vertical axis representing the rate of change of the reactor temperature, and uncontrollable states are hatched. In FIG. 14, states s21, s31, and s32 indicate uncontrollable states. Note that, since the variables used to define the states include the reactor temperature and the rate of change of the reactor temperature, as well as the jacket inlet temperature and motor torque, other variables may be used on the second axis. Alternatively, three variables may be selected and mapped on a three-dimensional graph.

[0075] As described above, by constructing a control model for controlling a plant in accordance with the changing characteristics of the plant, the control device can continue operation as much as possible even if, for example, a device in the plant breaks down and the characteristics of the plant become different from past characteristics. Below, we will explain the operation when the control device is actually used in operation.

[0076] FIG. 15 shows a configuration example of an online control system 1500 when controlling a plant 1 online using this system. This system includes a plant 1, a control device 5, controllability information 8, and a controllability information display unit 9. The plant 1, controllability information 8, and controllability information display unit 9 have the same configurations as those already described. In addition to the functions already described, the control device 5 receives operating data from the plant 1 to be controlled, including measurement values ​​(e.g., reactor temperature Tr, jacket inlet temperature Tci, and cooling water flow rate Fc). A state determination unit 51 of the control device 5 determines the characteristics and plant state of the plant 1 to be controlled based on the measurement values ​​included in the operating data received from the plant 1, using the method described above. At this time, the state determination unit 51 imports the controllability information 8 corresponding to the operating data assigned by the control system 1000. The state determination unit 51 determines whether the measurement values ​​of the operating data corresponding to the imported controllability information 8 and the measurement values ​​of the operating data received from the plant 1 satisfy a predetermined relationship (e.g., a condition that both are within a certain range). Then, when the state determination unit 51 determines that both of the above are within a certain range, it determines that the acquired controllability information 8 is applicable to the operation data received from the plant 1. When the acquired controllability information 8 is "not controllable," the state determination unit 51 issues an alarm to the plant 1. At that time, the controllability information display unit 9 can also display what plant state the plant 1 has entered on a screen such as that shown in FIG.

[0077] On the other hand, if the acquired controllability information 8 is "controllable," the state determination unit 51 operates the plant by outputting optimal manipulated variables using the control model 53, since safe operation is possible. The control model 53 is a control model created or updated by the method already described (for example, the method shown in FIG. 2).

[0078] FIG. 16 is a diagram illustrating an example of the configuration of a linkage system 1600 in which the control system 1000 shown in FIG. 1 and the online control system 1500 shown in FIG. 15 are linked together. The online control system 1500 in the linkage system 1600 receives the above-described controllability information 8 from the control system 1000. A state determination unit 51b, similar to the state determination unit 51 shown in FIG. 1, compares the above-described controllability information 8 with the measured values ​​of the operation data 4 to determine whether a predetermined relationship is satisfied, thereby determining whether the plant state of the plant 1 is normal or abnormal. The processing performed by the state determination unit 51b has already been described, and therefore its description will be omitted here. The state determination unit 51b reflects the result of the above determination in a state transition model 52b, similar to the state transition model 52, and creates a control model 53b, similar to the control model 53.

[0079] A control device 5a similar to the control device 5 creates a state transition model 52b and a control model 53b that reflect the above-described performance. As described with reference to FIG. 2, a simulator 2 executes a simulation as necessary based on the evaluation results by the control performance evaluation unit 6 and the operating conditions set by the condition setting unit 3. The simulator 2 then accumulates the pseudo-operation data obtained by the simulation as operating data 4. The accumulated operating data 4 is input to the control device 5a, which creates a state transition model 52a and a control model 53a that incorporate the evaluation results and the operating conditions as feedback. Furthermore, a controllability information adding unit 7 outputs controllability information 8 corresponding to the labels assigned in association with each state of the state transition model 52 to a controllability information display unit 9 and also to an online control system 1500, based on the evaluation results by the control performance evaluation unit 6.

[0080] In the online control system 1500, as described in FIG. 15 , the state determination unit 51b, upon receiving the controllability information 8, determines whether the plant 1 is currently in a controllable plant state based on the operation data 4 received from the plant 1 and the controllability information 8. This determination has already been described, so its description will be omitted here. If the state determination unit 51b determines that the plant 1 is in an uncontrollable plant state, it issues an alarm to the plant 1. Furthermore, if the state determination unit 51b determines that the plant 1 is controllable, it outputs optimal manipulated variables using the control model 53, as described above, to operate the plant. The state transition model 52b and the control model 53b used here are the latest models of the state transition model 52a and the control model 53a created in the control system 1000 and migrated 1601 to the online control system 1500 at a predetermined timing.

[0081] A control device 5b, similar to the control device 5 of the online control system 1500, controls the plant 1 using manipulated variables obtained using these models evaluated by the control system 1000. Furthermore, operational data of the plant 1 when controlled by the online control system 1500 is input to the control device 5b and fed back as operational data 4 input to the control system 1000. By linking the control system 1000 and the online control system 1500 in this manner, the results of the offline evaluation of the plant 1 can be fed back to the online control system 1500, and appropriate control based on the evaluation results can be performed in real time on the plant 1 to be controlled. In addition, in this embodiment, the control model 53a is created based on a state transition model 52a that comprehensively defines the relationship between the states and operations of the plant 1, and the optimal operation for achieving the target state is specified for each state. Therefore, control in response to changes in the characteristics of the plant to be controlled is possible using the control model 53b migrated from the control model 53a, in which a structure (i.e., state definition) that affects the control performance is appropriately set.

[0082] As described above, by using the control device of this embodiment, even if a plant device breaks down and the characteristics of the plant change from their previous characteristics, it is possible to continue operation as much as possible, and if continuing operation would cause problems, it is possible to stop operation and immediately issue an alarm.

[0083] Specifically, as described in FIGS. 1, 15, 16, etc., in a control system (online control system 1500) that controls a plant (e.g., plant 1) to be controlled by a computer 1600 having a processor (CPU 1602) and a memory (memory 1602), the computer defines a second state to be transitioned by an action in a first state as an action (e.g., an optimal action) that satisfies a predetermined condition among the actions in a state transition model (e.g., state transition models 52, 52a, 52b) defined for each characteristic of the plant. A control model (e.g., control models 53, 53a, 53b) is stored in the memory for each characteristic of the plant, and the processor determines the state of the plant for the characteristics based on operation data (e.g., operation data 4) obtained from the plant and the state transition model (e.g., the state determination unit 51b executes the process), and outputs operation information for operating the plant in the determined state for the characteristics based on the control model and the control model (e.g., the control device 5b executes the process). Thus, even if the characteristics of the plant to be controlled change, the plant can be controlled in accordance with the change.

[0084] 1, 2, etc., the control system includes a simulator (e.g., simulator 2) that executes a simulation of a plant with certain characteristics based on the control model and the plant's operating conditions (e.g., operating conditions set by condition setting unit 3) determined in accordance with the evaluation results of the plant's control performance, and outputs pseudo-operation data (e.g., operation data 4). The processor evaluates the plant's control performance based on the control model and predetermined constraints on the plant and outputs the evaluation results (e.g., control performance evaluation unit 6 executes this process), determines the plant's state using the operation data including the simulation data (state determination unit 51a executes this process), and creates the control model for each characteristic of the plant from the state transition model including the determined plant state. This allows the control model to be created using not only actual operation data but also pseudo-operation data that satisfies the constraints, thereby enabling accurate plant control.

[0085] 2, 6, 7, etc., the state transition model is a model determined based on the probability of transition from the classified first state to the second state, in which the plant state is classified by a predetermined method (for example, clustering technology) using the operation data, and the processor stores in the memory the control model that defines the action with the highest probability from the state transition model as the action that satisfies the predetermined condition, and outputs the operation information in the state based on the determined plant state and the control model. As a result, the state transition model is created based on the action with the highest probability among the actions that can be taken between states, making it possible to control the plant in a manner that is more suited to actual operation.

[0086] 10, 11, etc., the processor adds controllability information (e.g., controllability information 8) indicating whether or not the plant is controllable to each state defined in the state transition model based on the state transition model and the evaluation result of the control performance of the plant (for example, the controllability information adding unit 7 executes this process). This makes it possible to determine whether or not the plant is controllable depending on each state of the plant.

[0087] 15, 16, etc., when the operating data in the state of the plant corresponding to the controllability information and the operating data obtained from the plant satisfy a predetermined relationship (for example, when the controllability information 8 is "controllable"), the processor outputs the operation information based on the state of the plant from which the operating data was obtained and the control model (for example, the state determination unit 51b executes this process). This makes it possible to operate the plant quickly based on the optimal operation amount output by the control model.

[0088] 15, 16, etc., if the operating data in the state of the plant corresponding to the controllability information and the operating data obtained from the plant do not satisfy a predetermined relationship (for example, if the controllability information 8 is "uncontrollable"), the processor outputs an alarm for the plant in the state in which the operating data was obtained (for example, the state determination unit 51b executes this process). This makes it possible to immediately issue an alarm for the plant, and to promptly deal with any abnormalities in the plant.

[0089] 5 and other figures, the computer stores the control model in the memory for each of the characteristics representing the state of the plant caused by an abnormality that has occurred in the plant, and the processor uses the control model to output the operation information for the determined state of the plant. Furthermore, the abnormality that has occurred in the plant includes a deterioration state of the plant, and the processor uses the control model stored in the memory for each of the characteristics representing the deterioration state of the plant to output the operation information for the determined state of the plant. This makes it possible to perform an operation appropriate for the plant abnormality that is encountered by using control models according to various types and degrees of abnormality (e.g., small transmission surface deterioration).

[0090] As described above, according to this embodiment, even when the characteristics of the plant to be controlled change, it is possible to control the plant in accordance with the change in characteristics. As a result, the target value is reached more quickly than in the past, and the processing time required for appropriate control of the plant can be reduced. This contributes to reducing the amount of power used, and enables environmentally friendly plant control.

[0091] The present invention is not limited to the above-described embodiments as they are, and in the implementation stage, the components can be modified and embodied within the scope of the gist of the present invention, or multiple components disclosed in the above-described embodiments can be appropriately combined. [Explanation of symbols]

[0092] 1. Plant 2 Simulator 3 Condition setting section 4. Operational data 5, 5a, 5b control device 6 Control Performance Evaluation Section 7 Controllability information addition section 8 Controllability Information 9 Controllability information display section 11 Reaction tank 12 Jacket 13 Stirring blade 14 Pump 15 valves 16 Existing control device 17 Temperature controller 51, 51a, 51b State determination unit 52, 52a, 52b State Transition Model 53, 53a, 53b Control Model 1000 Control System 1500 Online Control System

Claims

1. A control system for controlling a plant to be controlled by a computer having a processor and a memory, The computer a control model defining an action that satisfies a predetermined condition among actions in a state transition model that defines a second state to be transitioned by an action in a first state for each characteristic of the plant, and storing the control model in the memory for each characteristic of the plant; The processor: determining a state of the plant in the characteristic based on operation data obtained from the plant and the state transition model; outputting operation information for operating the plant in the determined state based on the determined state of the plant in the characteristic and the control model; A control system comprising:

2. The control system includes: a simulator that executes a simulation of a plant having certain characteristics based on the control model and on operating conditions of the plant determined in accordance with an evaluation result of the control performance of the plant, and outputs pseudo-operation data of the plant; The processor: evaluating the control performance of the plant based on the control model and predetermined constraints on the plant and outputting the evaluation results; determining the state of the plant using the operational data including the simulation data; creating the control model for each characteristic of the plant from the state transition model including the determined state of the plant; 2. The control system of claim 1.

3. the state transition model is a model determined by classifying a state of the plant by a predetermined method using the operation data, and based on a probability of transition from the classified first state to the classified second state; The processor: storing in the memory the control model in which the action with the highest probability among the state transition models is defined as an action that satisfies the predetermined condition; outputting the operation information in the determined state of the plant based on the determined state and the control model; 2. The control system of claim 1.

4. The processor: adding controllability information indicating whether or not the plant is controllable to each state defined in the state transition model based on the state transition model and an evaluation result of the control performance of the plant; 3. The control system of claim 2.

5. The processor: outputting the operation information based on the state of the plant from which the operating data was obtained and the control model when the operating data in the state of the plant corresponding to the controllability information and the operating data obtained from the plant satisfy a predetermined relationship; 5. The control system of claim 4.

6. The processor: When the operation data in the state of the plant corresponding to the controllability information and the operation data obtained from the plant do not satisfy a predetermined relationship, an alarm is output for the plant in the state in which the operation data was obtained.

5. The control system of claim 4.

7. The computer storing the control model in the memory for each of the characteristics representing a state of the plant caused by an abnormality occurring in the plant; The processor: outputting the operation information in the determined state of the plant using the control model; 2. The control system of claim 1.

8. The abnormality occurring in the plant includes a deteriorated state of the plant, The processor: outputting the operation information in the determined state of the plant using the control model stored in the memory for each of the characteristics representing the deterioration state of the plant; 8. The control system of claim 7.

9. A control method for controlling a plant to be controlled by a computer having a processor and a memory, comprising: a control model defining an action that satisfies a predetermined condition among actions in a state transition model that defines a second state to be transitioned by an action in a first state for each characteristic of the plant, and storing the control model in the memory for each characteristic of the plant; determining a state of the plant in the characteristic based on operation data obtained from the plant and the state transition model; outputting operation information for operating the plant in the determined state based on the determined state of the plant in the characteristic and the control model; A control method comprising:

Images